Study On Virtual Network Embedding Algorithm Based On Delay Optimization

Network virtualization is considered to be an important means to solve the current rigid problem of traditional network architecture.Virtual network embedding is an important step in network virtualization.Its main research problem is how to provide as many network services as possible with as few network resources as possible.Therefore,the optimization goals of traditional virtual network embedding algorithms are usually to maximize the mapping benefit and minimize the mapping overhead.Nowadays,with the increasing number of delay-sensitive network applications on the Internet,the quality of service(QoS)of network applications,especially the average delay,is gradually valued by users.A virtual network embedding algorithm for delay-sensitive application environment can effectively improve the QoS of network applications.This thesis does the following research on network virtualization and virtual network embedding:1.Based on the review of the reading literatures,the virtual network embedding algorithms proposed by the academic circles in recent years are introduced,and the advantages and disadvantages of these algorithms are analyzed and summarized according to different classification methods.2.Aiming at the disadvantage that the time delay of the mapping result of the traditional virtual network mapping algorithm is not ideal,this thesis proposes a dynamic virtual network embedding algorithm based on node ranking.In the process of node ranking,the importance of network topology and resource parameters are comprehensively considered,and a parameter to measure the influence of network nodes on the average delay of mapping results is proposed.The node mapping process of the algorithm adopts the greedy algorithm;the link mapping adopts the combination of K-Shortest algorithm and priority.In addition,to ensure good performance in the dynamic mapping environment,the algorithm is optimized using a time window model and a remapping method.The simulation results show that the algorithm can improve the performance of mapping success rate,long-term mapping revenue-overhead ratio,and effectively improve the average delay of mapping results.3.This thesis proposes a virtual network embedding algorithm based on coarsening network requests.The algorithm uses a coarsening algorithm,which greatly simplifies the topology of the virtual network and reduces the difficulty of mapping through multiple rounds of iterations.In order to make the resource allocation more reasonable when the algorithm meets the network resource requirements and delay requirements of virtual network requests,the algorithm introduces a virtual network request classification algorithm to classify and process different types of virtual network requests.The simulation results show that the algorithm not only ensures the low average delay of the mapping results,but also improves the mapping success rate and resource utilization in delaysensitive application scenarios.